We propose to enhance the infrastructure of ARIC; the Atherosclerosis Risk in Communities, cohort to yield a new Cancer Epidemiology Cohort that brings novel features to cancer epidemiology research. In 1987, 15,792 participants aged 45-64 years were recruited from Forsyth Co., NC; Jackson, MS; Minneapolis, MN; and Washington Co., MD. 55% are women and 27% are African-American. Participants underwent 4 clinical exams; a 5th is scheduled for 2011. Blood and urine specimens have been banked, medications recorded, and a food frequency questionnaire completed. Participants were interviewed by phone annually and now semi-annually to obtain updated health information. The response at year 21 is 91%. Because ARIC has never been viewed as a Cancer Epidemiology Cohort and infrastructure and cost constraints, only cancer diagnosis has been systematically recorded. By 2006, 3,145 participants were diagnosed with an incident first primary and 376 with a 2nd / 3rd primary. Information, such as stage, grade, and histology, location in organ, laterality, receptor status, treatment, recurrence, and re-treatment, needed to address contemporary questions is not currently available. Tissue needed for molecular/genetic studies has not been collected. Thus, we propose: 1) Starting 2012, to prospectively identify cases from semi-annual phone interviews and collect medical/pathology records pertaining to cancer diagnosis, treatment, recurrence, and retreatment and tissue blocks. 2) To retrospectively collect information characterizing cancer diagnoses and recurrences before 2012 from cancer registries in the 4 ARIC states (consent already obtained) and medical records, and collect tissue blocks. By 2016, we expect 4,900 fully annotated incident cases. We established a Cancer Working Group to develop protocols for adjudicating cancer endpoints and prioritize research using the resource. With enhanced infrastructure, we expect that research questions such as these are addressable uniquely in ARIC: 1) What is the association between timing of the natural history of diabetes using 4 fasting glucose and 2 HbAlc measurements and timing of cancer diagnosis? 2) Using GWAS and sequencing data, is there a set of risk variants shared by major cancers or are variants specific to each site?